Active fault segmentation and seismic hazard in Hoa-Binh reservoir, Vietnam

Cent Eur J Geosci • 5(2) • 2013 • 223-235 DOI: 10.2478/s13533-012-0128-5 Central European Journal of Geosciences Active fault segmentation and seismic hazard in Hoa-Binh reservoir, Vietnam Research Article Phan Trong Trinh∗ , Hoang Quang Vinh, Nguyen Van Huong, Ngo Van Liem Institute of Geological Sciences, Vietnam Academy of Science and Technology, 18 Hoang Quoc Viet, Hanoi Received 12 March 2013; accepted 26 May 2013 Abstract: Based on remote sensing, geological data, geomorphologic analysis, and field observations, we determine the fault system which is a potential source of earthquakes in Hoa-Binh reservoir It is the sub-meridian fault system composed of fault segments located in the central part of the eastern and western flanks of the Quaternary HoaBinh Graben: the Hoa-Binh fault is east-dipping (75-80◦ ), N-S trending, km long, situated in the west of the Hoa-Binh Graben, and the Hoa-Binh is a west-dipping (75-80◦ ), N-S trending; 8.4 km long fault, situated in the east of the Hoa-Binh Graben The slip rate of normal fault in Hoa-Binh hydropower dam was estimated at 0.3-1.1 mm/yr The Maximum Credible Earthquake (MCE) and Peak Ground Acceleration (PGA) in the HoaBinh hydropower dam have been assessed The estimated MCE of HB.1 and HB.2 is 5.6 and 6.1 respectively, and the maximum PGA at Hoa-Binh dam is 0.30 g and 0.40 g, respectively The assessment of seismic hazard in HoaBinh reservoir is a typical example of seismic hazards of a large dam constructed in an area of low seismicity and lack of law of seismic attenuation Keywords: fault segment • normal fault • state of stress • maximum credible earthquake • peak ground acceleration ã seismic moment â Versita sp z o.o Introduction The Hoa-Binh hydropower dam, more than 120 m in height, is one of the highest dams in southeastern Asia This dam is also one of the largest dams constructed on uncompacted foundation in the world The power plant is of great importance to the economic development of the Vietnam However, in the Hoa-Binh area, two active normal faults are located less than 2.5 km from the dam to the east Therefore, the seismic hazard of the dam threatens directly not only Ha Noi city, but all the ∗ E-mail: phantrongt@yahoo.com; trinh-pt@igsvn.ac.vn Red River Delta area - with a population of more than 25 million To prevent the disaster and to ensure the absolute safety of the Hoa-Binh hydropower dam, we have carried out detailed analysis of geology, geomorphology, and remote sensing in the Hoa-Binh hydropower zone to make a seismic hazard assessment and proposals of the most sensible operation for the Hoa-Binh reservoir In areas of low density of seismicity with small deformation rate, the pure probability approach is less efficient where recording time is too short and is carried out in areas with long recurrence interval of earthquakes In this case, it is necessary to assess seismic hazards by using simultaneously seismotectonic methods and probability analysis for each site or for certain areas The seismogenic capability of active faults is identified through satellite images, to223 Active fault segmentation and seismic hazard in Hoa-Binh reservoir, Vietnam pographical and geological maps, field survey and earthquake catalogue The capability at a site is characterized by Maximum Credible Earthquake (MCE) and Peak Ground Acceleration (PGA) The short evaluation of Maximum Credible Earthquake in Hoa-Binh area and seismotectonics along the Red River fault zone was presented in [18] This paper presents in detail some new arguments of seismicity, induced earthquakes, state of stress, active fault segmentation and its slip rates This is the basis for the estimation of the Maximum Credible Earthquake We present also a new approach to estimate maximum PGA, earthquake magnitude and PGA corresponding to various returned periods by the combination of deterministic and probabilistic approaches This work is an example of the study of seismic hazards of a large dam constructed in an area of low seismicity Tectonic setting The collision between the India and Eurasia plates that took place 50 Ma ago has changed the tectonic framework of Asia [24, 30, 31] Systematic studies along the Red River Fault zone (RRFZ) from Yunnan to Vietnam [11, 12, 15, 20] proved that almost all gneiss structures situated along the Red River metamorphic zone were formed during Cenozoic Major left-lateral ductile shear occurred along this zone in the Oligo-Miocene as Indochina was extruded several hundred of kilometers towards the southeast [11, 15, 22] Neotectonic activity in the Northwest region has been expressed not only by the left lateral strike-slip displacement of the RRFZ, but also occurring along a series of fault zones in the same NW-SE direction [23, 24] One can observe a series of structures of the Northwest region that was formed during Cenozoic [9] Over thrusting phenomena were widely developed in the Northwest region as well as Northeast region, such as Hoa-Binh arc and Sa Pa marble [8, 15] In North Vietnam, the Red River fault zone splays into two major active fault branches The Red River branch, representing the SW limit of the Day Nui Con Voi metamorphic massif, follows the Red River valley [10] The Song Chay River branch, which is remarkably clear on Landsat and SPOT images and in its geomorphology, is located NE of Day Nui Con Voi Toward the SE, the Red River fault itself splays into several faults The two major ones are located NE and SW of the Red River Right lateral strikeslip offsets along these faults are determined by analyzing tributaries, stream channels, Quaternary alluvial fans, and river valleys on Landsat and SPOT images, on detailed topographical maps, and by field observations [1, 18, 34] The fresh geomorphologic appearance of the faults under 224 these intense erosive conditions demonstrates that these faults are active [18] At present, there is a contradiction between the active fault slip rate calculated from GPS and that calculated according to the geologic data in North Vietnam [1, 4, 5, 19, 31, 32, 34, 35] In the Cenozoic, the Hoa-Binh area is the result of extrusion of IndochinaSunderland block along the Red River fault zone The arc system formation probably results from a decrease in the amount of lateral movement in the Red River fault zone to the southeast [15] This caused the partially northwest – southeast compression leading to the formation of a large thrust zone with the thrust - trend from the northwest to the southeast Consequently, there are completely mylonized zones strongly foliated and folded on a large scale For example in the Doc Cun area, middle Permian – Early Triassic and middle Triassic formations were entirely mylonized and nearly have E-W foliation, gently slope angle dipping to the north or the northwest The foliation trend exhibits a large thrust nappe thanks to a compressive force directed from the northwest to the southeast After this period, it appeared to be an intense folding on a large scale, also on this site many limbs cropped out with a sub-parallel axial surface trend GPS data from 1994 to 2007 have shown that, in the present day, the southwestern and northeastern sides of the Red River Fault are moving eastward at the rate of 34.5 mm/yr and southward at a velocity of 12 mm/yr [4, 5, 25] The Hoa-Binh dam zone is the regional boundary between RRFZ and Da River zone (Fig 1) Uplift of the N-S fault system is expressed clearly by a high and average topographical elevation of more than 200 m in comparison with the southwest wing This fault is an active normal fault Typical hanging valleys prove the fast uplift of the NE wing that exceeds the erosion of gravel, sand and soil The Hoa-Binh N-S fault system does not extend continuously, but it is divided into short segments Strong separation and extensive activity of the fault are manifested in this segment; the fault system is divided into discontinuous segments that are called as Hoa-Binh and Hoa-Binh Fault segments Materials and Methods 3.1 Materials: In this area, we used geological maps of scale 1/50.000 and 1/200.000 from the Department of Geology and Mineral Resources, with air photos and topographic maps of scale 1/50.000 from the Department of Geodesy LANDSAT and SPOT satellite images were used for the investigation of active faults The earthquake catalog employed Phan Trong Trinh et al from the ith fault on R* becomes less weighty when it is different from R* 4) The stress tensor is determined by minimizing the deviation between the parameter Rj and the representative value R* 5) The criterion of fracture is taken into account 6) The influence of fault planes on the deviation of stress ratios is taken into account The stress tensor is calculated by minimizing via least squares, the following function: N F= i=1 [Ri (α, β, γ, ni , si ) − R ∗ ]2 exp Figure Sketch of active faults in North Vietnam is from the Institute of Geophysics, Vietnam Academy of Science and Technology In the field, we investigated geomorphology of active fault and profiles of radon investigations We performed geological investigations along cross sections by trenching 3.2 Methods: 3.2.1 State of stress State of stress is an important parameter in the evaluation of the Maximum Credible Earthquake From fault populations, one has information on normal vectors of the faults and slip vectors of the striations on the surface of the faults The slip on fault surface is the result of different tectonic phases One has to determine different stress tensors corresponding to various tectonic phases One pays more attention to stress tensor of latest tectonic phase or recent tectonic phase related to earthquake source The inverse method used to separate tectonic phases [16] is as follows: 1) An intermediate parameter is defined, which is a piecewise linear function of the aspect stress ratio Each value of this parameter is assumed to have its own probability density 2) A value R*, corresponding to the maximum of the probability density function, is taken as a representative value for the determination of the stress ratio 3) The influence of the parameter Ri determined [Ri (α, β, γ, ni , si ) − R ∗ ]2 2σc α, β, γ are Euler angles ni , si are ith normal and slip vectors deduced from ith fault In general, we observe a polyphase superposition of stress field in the studied region The principle used to separate the different tectonic phases is as follows All the faults are used to determine the function F At the first stage of calculation, the stress tensor corresponding to the global minimum of F is considered to be that of the latest tectonic phase A population of faults defining this tectonic phase is eliminated before determining the new function F in the second stage of calculation The stress tensor corresponding to the second tectonic phase is obtained by minimizing F The processing continues until the remaining fault set cannot define a stress tensor 3.2.2 Active faults One considers a fault as an active fault if one has an evidence of displacement or paleoearthquakes during the Holocene In the area of low deformation, Pleistocene faults are those active faults that have been investigated at the surface and which has evidence of movement in the past 1.6 million years From satellite images and aerial photo, one can reveal an active fault from a fault scarp The fault scarp is the topographic expression of fault related to the displacement of the surface by slip along faults It is a small step on the land surface where one side of a fault has moved with respect to another They are a result of various movements and erosion along an old fault or by a slip on a recent fault Various methods of remote sensing such as directional filtrate, the integration of digital elevation modelling (DEM), and satellite images can help us to identify fault scarp Fault scarp is usually the representative of normal fault or strike-slip fault Due to the uplift along the fault, the fault scarp is particularly susceptible to erosion, especially if the material being uplifted consists of unconsolidated sediment The geomorphological observation in the field can reveal if it is active or old fault Fault scarp that cuts across Holocene fan 225 Active fault segmentation and seismic hazard in Hoa-Binh reservoir, Vietnam is a candidate of active fault One can recognize normal active fault from the observation of triangular facet along fault scarp The segmentation of active fault is affected generally by the observation of fault scarp length Along fault scarp, the representation of triangular facets is an argument of normal active fault One can decide to make trenching across fault scarp, in the Holocene fan The evidence of displacement along faults in Holocene is a direct indicator of active fault If one has the offset of a sediment layer at two sides of the fault and the age of this sediment, one can estimate of slip rate of active fault From trenching across fault scarp, one can reveal events of paleoearthquake in the pass The age of Holocene sediment is dated generally by isotope Carbon 14 3.2.3 Radon investigation Radon is a naturally occurring radioactive gas that results from the decay of trace amounts of Uranium and Thorium, found in most rock and soil Radon gas levels vary from location to location It is possible for any granite to contain varying concentrations of uranium and other naturally occurring radioactive elements These elements can emit radiation and produce radon gas, a source of alpha and beta particles and gamma rays Some types of granite may emit gamma radiation above typical background levels The Uranium stays in the ground, but Radon gas seeps upward along active faults One uses Alpha-track radon gas detectors to investigate radon gas along crosssection, across the fault scarp Each Alpha-track radon gas detector contains a detector element, called a foil When radon atoms decay inside the detector, they release alpha particles If the alpha particles strike the foil, they make microscopic tracks in the surface of the foil When the detector is analyzed, the foil is chemically treated to enlarge the alpha tracks, which are counted on an automated system The average radon level is calculated from the number of tracks, and the number of days the detector was exposed Radon concentration is usually measured in Becquerel per cubic meter (Bq/m3 ) Typical domestic exposures are about 100 Bq/m3 indoors, and 10–20 Bq/m3 outdoors It is measured in picocuries per liter (pCi/L) in the USA, with pCi/L=37 Bq/m3 Radon levels vary with daily and in response to weather conditions and ventilation patterns For these reasons, it is preferable to make the measurements for a long period of days For each region, one establishes a map of the background level of Radon One considers the existence of an anomaly of radon if the radon level is times higher than background levels Radon investigation is only indirect method to detect active faults These techniques are only used as a complementary method 226 3.2.4 Maximum Credible Earthquake One uses fault segments to assess Maximum Credible Earthquake (MCE) by seismotectonic methods, integrating many formulations according to fault length, rupture area and seismic moment [3, 4, 14, 16, 19] One can give an example of the determination MCE from the fault area according to Well – Corpersmith [26], regressions of Surface Regression of Surface Rupture Length and Moment Magnitude for the state of stress of extension is following: M = 4.86 + 1.32 ∗ log(SRL) (1) M is moment magnitude; SRL is Surface Rupture Length; standard errors of a and b are 0.34 and 0.26, respectively Regression of Rupture Area, and Moment Magnitude for the state of stress of extension [26] is M = 3.93 + 1.02 ∗ log(RA) (2) RA is Rupture Area; standard error of a and b are 0.23 and 0.10, respectively Regression of Moment Magnitude (M) and Maximum Displacement for state of stress of extension [26] is M = 6.61 + 0.71 ∗ log(MD) (3) MD is maximum displacement; standard error of a and b are 0.09 and 0.15, respectively On seismic moment approach [6], a physically meaningful link between earthquake size and fault rupture parameters is seismic moment; the seismic moment is determined by: Mo = µDLW (4) Where Mo is seismic moment, is the shear modulus, 3.0ì1011 dyne/cm2 for crustal faults [6], L is rupture length, W is rupture width, and D is the maximum slip displacement MCA is calculated from [6]: M = 2/3 ∗ log(Mo) − 10.7 (5) where M is MCA The amount of maximum displacement is unknown Nonetheless, it could be estimated by continuous approximate solution deduced by a formula of maximum displacement according to Well-Coppersmith [26] D is deduced from and Calculating of MCE is an iterative process In the first step, the amount of maximum displacement is evaluated based on data acquired from assessing maximum earthquake according to different approaches, from which to estimate seismic moment Phan Trong Trinh et al After averaging main numbers and normalizing errors, the amount is deduced from Well-Coppersmith’s formula In the second step, based on seismic moment, the magnitude of maximum earthquake is determined, and this process is iteratively operated until acquired results are stable [18] To calculate MCE, the fault depth can be estimated in ways The first is based on the depth of earthquake hypocenter in the area and along the fault zone The second is based on the change in mechanical behavior of the earth’s crust with depth In our study, the methods used to calculate MCE are based on the state of stress combined with the fault length [16, 19] and the fault area from WellCoppersmith [26], Wyss [29], Woodward-Clyde [28] We take the coefficient for fault length approach because the information deduced from fault length is less confidence, coefficient for rupture area approach because the information deduced from fault area is high reliability and coefficient for seismic moment due to its most physical signification 3.2.5 Seismic hazard We use the attenuation models 1, 2, 3, of Campbell [3], Idriss [7], Xiang & Gao [33], Woodward-Clyde [28], Ambraseys [2] For example, the Xiang & Gao [33] formulation is as follows: PGA = 252.9e0.5155 (R + 10)−1.1516 where R (Km), is the distance from earthquake hypocenter to the site One takes the weighted average to calculate PGA by different methods The model of Campbell is based on global data of strong ground motion near the source, so it has high reliability in the case of assessing earthquakes within 50 km or less The above formulae can use the coefficient for calculating the average weight The formula of Xiang & Gao [33] can use the coefficient because it is set up from data of earthquake in Yunnan region, which has a similar geological setting and structure as that in Vietnam For estimating Magnitude and Peak Ground Acceleration corresponding to return periods, we use a b value of 0.7 in GutenbergRicher line determined by Nguyen [13] The probabilistic approach usually uses the seismotectonic model as some area with different MCE In our case, we use a narrow area corresponding to fault segmentation with MCE determined by deterministic approach and surrounding area of no earthquake The calculation of ordinary probabilistic approach with this simple model will give us a combination of deterministic and probabilistic approaches Figure Seismotectonic map of Hoa-Binh hydropower dam area (modified from Phan Trong et al., 2012) Results 4.1 State of stress Using the inverse method [16], we separate tectonic phases from fault population The stress tensor of the latest tectonic phase corresponding to the solution chosen from 70% fault population This is an extension with maximum stress axis σ1 of sub-vertical The minimum stress axis σ3 is sub-horizontal with the direction E-W This tectonic phase of Pliocene – present time is observed along the Red River fault and Northwestern region [18] The older tectonic phase corresponding to 30% fault population has intermediate stress axis σ2 of sub-vertical and minimum stress axis of sub-meridian The distribution of the two stress fields is represented in the Figure The state of stress corresponding to the present time has significations; it is used to compare with the kinematics of active fault, and to estimate MCE 4.2 Radon anomaly We carry out profiles of Radon investigation across the fault scarp in the Hoa Binh region The cross-sections prolong 100-150 meters with the distance of each foil of meters The background level of Radon in Hoa Binh area is 25 Bq/m3 At some site next to fault scarp, one can observe a Radon anomaly of 70 – 85 Bq/m3 This anomaly is relatively high characterizing the activity of 227 Active fault segmentation and seismic hazard in Hoa-Binh reservoir, Vietnam Figure Figure Active fault system in Hoa-Binh hydropower dam the active fault However, Radon anomaly are not typical along the fault scarp At some sites of fault scarp, the Radon level is only equal to the background 4.3 Active fault segmentation From Landsat, Spot satellite images and DEM, we defined three N-S trending segments developed next to the Hoa-Binh dam root which generated a nearly verticalflank graben (Fig 3) In some localities, the graben is 2.5 km wide and files up with alluvial-pluvial deposits of up to 70 m in thickness The fault segment in the eastern flank is 8.4 km long, and that in the west flank is km long The distance between the latter and the dam is only 0.3-2.5 km Extensive displacement of the two fault segments can be clearly observed in Spot images, Landsat images, DEM and in the field This movement is apparently characterized by triangular facets (Fig 4) The result obtained from geomorphological observation is entirely consistent with the stress field that is predominated by E-W extension in Pliocene-Quaternary [17] We identified the fault systems which are potential earthquake sources In the trench, one can observe mylonite and fracture zones The sub-meridian fault system consists of segments located in the central part of the eastern and the western flanks of the Quaternary Hoa-Binh graben Aside from faults controlling the Hoa-Binh graben - which have length of more than 10 km - other faults often are less In the contemthan 10 km long (Fig and Fig 4) 228 Stereo model of Landsat image showing the normal active fault occurring near the Hoa-Binh hydropower dam poraneous tectonic framework, this is the youngest fault cutting all older structures and former framework On each flank of the Hoa-Binh - Bat Bat graben, the normal fault planes are nearly 60-65◦ , dipping towards the graben center, and each fault coincides with some landforms like fault scarp or triangular facets and controls the distribution of river terraces and alluvial layer In a general view, all of the regions belonging to the SW part of the RRFZ, Quaternary grabens and tectonic breccia zones developed close to the Red River Fault with acute angle exhibit right lateral movement regime of the RRFZ itself during Late Cenozoic The sub-meridian fault system controls the structure of the Hoa-Binh graben and is distributed along its both flanks The western fault branch running across the Ong Tuong Hill area is more than km long, with the fault plane dipping eastwards Along this segment, it was recognized a number of normal active shear zones in the trenching across the fault scarp The eastern fault branch is formed by one segment whose length is about 8.4 km (Fig 3) Along this segment, triangular facets characteristic for normal faulting markedly developed In addition, the active shear zone system in the Doc Cun area is also well identified The fault controls the river flow and stream system according to different base levels The Da river changes direction from WNW-ESE to N-S, following the Hoa Binh graben, before confluence with the Red River Many ponds and swamps are located along the graben Fault systems can be seen clearly on satellite images and in actual topography through fault scarps and triangular facets Remarkably, along some segments in the east of the Hoa-Binh depression, we identified a series of triangular facets in Lang Ngoi and Lang Su areas (Fig 4) These facets have a height of 70-100 m and the width of foot side of more than 500 m Along with some segments in the west of the Hoa-Binh depression, such as in the Ong Tuong Hill and Doc Cun areas, the active shear zones are of the normal slip Fault gauges consist of clay, debris of ferrous-gel in the fine-soft-porous state In some sites, one can ob- Phan Trong Trinh et al Figure Active shear surface system cutting across the bedding plane system in Triassic siltstones in the Ong Tuong Hill area, Hoa-Binh City (shear surface, slickensides and bedding plane of E-W trend and nearly vertical slope angles) filled up with black-gray or yellow-brown or red-brown clay materials B Triangular facets of normal fault in east Bai Yen, Hoa-Binh serve striations on the fault plane In Ong Tuong hill area, the active shear zones cut across 20-50 m-wide bedding plane system (bedding plane of E-W trend, dip of 7580◦ ), and they are filled up with black gray clay material (Fig 5) Landslides and cracks occurred in 1996 without earthquake along fault scarp in the eastern branch of the Hoa-Binh graben In the Doc Cun area and low hill range situated in the east of the Hoa-Binh dam, one can observe a shear zone of larger 60-80 cm The slip on the shear zone gives evidence of extensive faulting 4.4 Fault slip rate The Hoa-Binh graben is filled up with formations of alluvial-colluvial pebble, gravel, sand and a little of marshy facies in the south of Hoa-Binh City The thickness of late Pleistocene sedimentary layers varies from site to the site and reaches 50-80 m in Hoa-Binh City From the age of late Pleistocene sediments (about 120 – 150 Ka), we estimate the fault slip rate as 0.3-0.7 mm/yr From the facet’s height of 60-100 m and summing, the facet’s age is related to Riss glacial cycle which occurred 55,000-150,000 years ago [18], we estimate the slip rate in order of 0.4-1.1 mm/yr Combining the two approaches, the rate of normal faulting in Hoa-Binh hydropower dam is estimated as 0.3-1.1 mm/yr 4.5 Maximum Credible Earthquake and seismic hazards There are little seismic and paleoseismic data, and seismic station network is less dense so no calculation on exact earthquakes location and focal mechanisms are available Earthquake catalog is mainly based on oral statistics and ancient documents from the 12th century to recent (Fig and Appendices) According to these documents, in that period there were several earthquakes with M>3 Figure Earthquake distribution in Hoa-Binh region and surrounding area Notably, after Hoa-Binh hydropower plant came into operation in 1989, the water level of the reservoir was raised, amd several induced earthquakes with M>3.0 occurred with seismogenic source from tectonic components of HoaBinh graben (Fig and Table 1) In order to assess seismic hazards in Hoa-Binh hydropower dam area, we focused on the fault systems of significant size The submeridian fault system is distributed along two flanks of the graben The western fault branch is more than km long with nearly vertical fault plane dipping to the east The shortest distance from the section Hoa-Binh (HB1) to Hoa-Binh dam is about 0.3 km, the fault dip is determined as 75◦ The eastern fault branch forms one segment of 8.4 km in length The shortest distance from the section Hoa-Binh (HB2) to the Hoa-Binh dam is about 2.5 km, the fault dip is determined as 70◦ For seismic hazard assessment, we estimated firstly the MCE From MCE and distance from the fault to the hydropower dam, we estimated the Peak Ground Acceleration (PGA) The estimation of MCE of Hoa-Binh fault segments in the Hoa-Binh dam is presented in the Table MCE of HoaBinh fault segment is 5.6 PGA caused by Hoa-Binh fault segment at the Hoa-Binh dam is represented in the Table Using the value b of 0.7 determined from seismic data [13], we estimate PGA corresponding to different recurrences by probabilistic approach Ground accelerations corresponding to return periods of 950 and 475 years are represented in the Table and the Table 5, respectively 229 Active fault segmentation and seismic hazard in Hoa-Binh reservoir, Vietnam Table Earthquake occurred in the lake and surroundings when the impounding of the Hoa Binh reservoir No Year Month Day Hour Minute second longititude latitude Depth (Km) M 1989 04 14 12 37 44.3 21.71 105.23 3.8 1989 05 14 13 37 47 20.82 105.36 - 3.8 45.7 20.82 105.38 1989 1989 1989 1989 1989 04 05 05 05 05 14 22 26 26 29 12 23 18 21 16 46 53.8 05 23.1 09 26.7 28 45 3.4 21.71 20.81 20.81 2081 Table 105.23 105.30 6-7 105.30 105.30 3-4 - 3.7 4.9 2.0 4.0 1.8 MCE and rate of displacement of Hoa Binh fault estimated from various models Input data are fault length of 8.4 km, fault depth of km, dip angle of 70, state of stress of extension Models Model of Slemmons,1982 for normal fault length MCE 6.1 Model of Well-coppersmith, 1994 for normal fault length 6.1 Model of Well-coppersmith, 1994 for normal fault area 5.7 Model of Woodward-clyde,1983 for fault area 5.8 Displacement (m) of fault segment, Slemmons, 1982 0.10 Maximum displacement (m) , Well-Coppersmith, 1994 0.15 Model of Wyss, 1979 for fault area Model of seismic moment, Hanks- Kanamori 5.9 5.2 Slip rate (mm) of fault segment, Woodward-clyde, 1983 0.07 Figure Induced earthquakes in 1989 when Hoa-Binh hydropower plant came into operation Discussion During the year 1970, in preparation for the construction of the Hoa-Binh dam, Vietnamese and Russian workers carried out a series of geological and geophysical investigations However, they did not pay attention to active faults For seismic hazard assessment, they used only the old experimental law The Cho Bo fault of the direction EW was considered as the source of earthquake occurred in this area However, in our study, we have not found any evidence of active displacement along Cho Bo fault The Hoa-Binh hydropower dam has been considered in the design of seismic intensity of on MSK scale It is difficult to compare the MSK intensity to the PGA in this study However, seismic hazard determined from this study is much larger than that one estimated by Russian experts [13] Using the first and third Gumbel’s asymptotic distribution, Nguyen estimated MCE for Vietnam from the earthquake catalog for the period 1903–1988 [14] The highest value of the maximum earthquake of magnitude 230 Average displacement (m), Well-Coppersmith,1994 0.14 is estimated for the Ma river zone within a return period of 123 years For North Vietnam, the occurrence of earthquakes with magnitudes higher than 6.0 within 50 years is predicted with 90% probability In the region where relative movements are significant and reach several cm/yr, seismic activity is intense enough to give a complete representation of the deformation in a short time However, in the Hoa-Binh region as in the whole Red River fault zone, the fault slip rates are limited to several mm/yr [18] In this case, earthquake recurrence is extremely long, and seismicity does not reflect the accumulated strain That is why our approach shows that it is more advantageous than the pure probabilistic one PGA was estimated for the first time in Vietnam in UNDP’s Project for assessment of earthquake hazard for the Hoa-Binh hydropower dam in 1993 [27] The evaluation of MCE was based only on the fault length [21] PGA of not greater than 0.4 g was estimated from the first model of Campbell Using more detailed investigations and more robust methods, the PGA determined in this study is more precise than that one determined from Phan Trong Trinh et al Table Peak Ground Acceleration provoked by segment Hoa Binh at Hoa Binh dam Model PGA (g) PGA calculated from model of Campbell (with M>6) 0.27 PGA calculated from model of Campbell, near field, 1997 0.41 PGA calculated from model of Campbell PGA calculated from model of Campbell, near field earthquake, 1988 PGA calculated from model of Idriss , 1982 0.37 0.24 PGA calculated from model of Cornell, 1979 ( reference only ) 0.34 PGA calculated from model of Ambraseys, 1995 PGA calculated from model of McGuire, 1980: ( reference only ) PGA from model of Estena-Rosenblueth, 1974:( reference only ) Weighted PGA (g) 0.28 0.44 0.70 0.36 0.4 ±0.07 Ground Acceleration with the recurrence interval of 950 year Model PGA calculated from model of Campbell PGA calculated from model of Campbell (with M>6) PGA calculated from model of Campbell, near field earthquake, 1988 GA (g) 0.20 0.23 0.25 PGA calculated from model of Campbell, near field, 1997 0.32 PGA calculated from model of Xiang J.-Gao Dong, 1989 0.20 PGA calculated from model of Idriss , 1982 PGA calculated from model of Woodward - clyde, 1983 PGA calculated from model of Ambraseys, 1995 0.32 0.23 0.36 PGA calculated from model of McGuire, 1980: ( reference only ) PGA calculated from model of Cornell, 1979 ( reference only ) 0.51 Weighted GA 0.31 ±0.07 PGA from model of Estena-Rosenblueth, 1974:( reference only ) Table 0.30 PGA calculated from model of Xiang J.-Gao Dong, 1989 PGA calculated from model of Woodward - clyde, 1983 Table 0.25 0.25 0.27 Ground Acceleration with the recurrence interval of 475 year Model PGA calculated from model of Campbell PGA calculated from model of Campbell (with M>6) PGA calculated from model of Campbell, near field earthquake, 1988 GA (g) 0.16 0.19 0.21 PGA calculated from model of Campbell, near field, 1997 0.26 PGA calculated from model of Xiang J.-Gao Dong, 1989 0.17 PGA calculated from model of Idriss , 1982 PGA calculated from model of Woodward - clyde, 1983 PGA calculated from model of Ambraseys, 1995 0.26 0.19 0.29 PGA calculated from model of McGuire, 1980: ( reference only ) PGA calculated from model of Cornell, 1979 ( reference only ) 0.37 Weighted Ground Acceleration 0.25 ±0.05 PGA from model of Estena-Rosenblueth, 1974:( reference only ) 0.18 0.20 231 Active fault segmentation and seismic hazard in Hoa-Binh reservoir, Vietnam Winter et al [27] The result from Winter et al based on only formula only does not permit an estimation of error [4] Conclusions Based on the air photos, satellite images, and geomorphologic observation, we identified the N-S active fault systems as potential sources of earthquakes in Hoa-Binh reservoir These fault systems consist of two fault segments: the first segment is east-dipping, km long, with the dip of 75-80◦ , and the second one is 8.4 km long, dipping westward with an angle of 70-75◦ We estimated the rate of normal faulting in Hoa-Binh hydropower dam as 0.3- 1.1 mm/yr The fault segment HB1 could produce the Maximum Credible Earthquake of 5.6 and Peak Ground Acceleration at Hoa-Binh hydropower dam of 0.3 g The fault segment HB2 could produce the Maximum Credible Earthquake of 6.1 and PGA at the Hoa-Binh dam of 0.4 g We suggest that, for evaluation of seismic hazard, we have to use various seismic, geological and geomorphological data and combination of different methods in the estimation of Maximum Credible Earthquake and Peak Ground Acceleration We also require a combination of deterministic and probabilistic approaches The assessment of seismic hazard in Hoa-Binh reservoir is a good example of the study of seismic hazards of a large reservoir constructed in an area of low seismicity, and the limitations of the seismic attenuation law [5] [6] [7] [8] [9] [10] [11] Acknowledgement We thank Leloup, Winter, Findlay and Ozer for useful discussions and field work assistances Financial support by NAFOSTED is gratefully acknowledged [12] References [1] Allen C R., Gillespie A R., Yuan H S., Kerry E., Buchun Z., Chengnan Z., Red River and associated faults, Yunnan Province, China; Quaternary geology, slip rates, and seismic hazard Geol Soc Am Bull., 1984, 95, 686–700 [2] Ambraseys N N., The Prediction of Earthquake Peak Ground Acceleration in Europe Earthquake Engineering and Structural Dynamics, 1995, 24, 467-490 [3] Campbell K W., Empirical near source attenuation relationships for horizontal and vertical components of peak ground acceleration, peak ground velocity and 232 [13] [14] [15] [16] pseudo absolute acceleration response spectra Seismological Research Letters, 1997, 68, 154-179 Duong C C , Feigle J., Geodetic measurement of Horizontal Strain across the Red River fault near Thac Ba, Vietnam, 1963–1994 J Geodesy, 1999, 73, 298– 310 Duong C C., Feigl K., Nguyen T Y., Tran D T., Phan Trong T., Pelzer G., Lacassin R., Tapponnier P., Dang H V., Tran D L., Terrestrial surveying measurement of deformation across the Red river fault near Thac Ba, Vietnam, 1963–1983 Proceed Inter Seis Haz South Asia, 1994, 34–39 Hanks T.C., Kanamori H., A moment magnitude Scale J Geophys Res., 1979, 82, 2981-2987 Idriss I.M., Evaluating seismic risk in engineering practice Proceeding of the 11th International conference on soil mechanics and foundation engineering San Francisco, 1985, Vol 1, 255-320 Findlay R., Phan Trong T., The structural setting of Song Ma region, Vietnam and the Indochina - South China plate boundary problem Gondwana Res., 1997, 1, 11-33 Lacassin R., Leloup H., Phan Trong T., Tapponnier P., Unconformity of red sandstones in north vietnam: field evidence for Indosinian orogeny in northern Indochina? Terra Nova, 1998, 10, 106-111 Lacassin P., Tapponnier H., Leloup T., Phan Trong N., Yem T., Morphotectonic evidence for active movement along the Red river fault zon, Proceed Inter Seis Haz South Asia, 1994, 66–71 Leloup P H., Lacassin R., Tapponnier P., Scharer U., Zhong D., Liu X., Zhangshan J., Phan Trong T., The Ailao Shan - Red river shear zone (Yunnan, China), Tertiary transform boundary of Indochina Tectonophysics, 1995, 251, 3-84 Leloup P H., Arnau N., Lacassin R., Kienast J R., Harrison T M., Phan Trong T., Replumaz A., Tapponnier T., New constraints on the structure, thermochronology and timing of the Ailao Shan – Red river shear zone, SE Asia J Geophys Res., 2001, 106, 6657-6671 Nguyen D X., Thanh T T M., Estimation of recurrence period of large carthquakes in a given place following the long - term variation of seismic activity J Geology (Hanoi) serie B, 1996, 7-8, 115-120 Nguyen H P., Probabilistic assessment of earthquake hazard in Vietnam based on seismotectonic regionalization Tectonophysics,1991, 198, 81–93 Phan Trong T., Cenozoic great overthrust in Hoa-Binh and its surrounding area J Geology (Hanoi) serie A, 1993, 214-215, 30-35 Phan Trong T., An inverse problem for the determi- Phan Trong Trinh et al [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] nation of the stress tensor from polyphased fault sets and earthquake focal mechanisms Tectonophysics, 1993, 224, 393-41 Phan Trong T., Tan M T., Liem N V., Vinh H Q., Huong N V., The D V., Closson D., Ozer A., Apport de la télédétection ŕ l’étude des failles actives et des risques sismiques au nord du Viet-nam J Télédétection, 2007, 7, 465-481 Phan Trong T., Liem N V., Huong N V., Vinh H Q., Thom B V., Thao B T., Tan M T., Ng H., Late Quaternary tectonics and seismotectonics along the Red River fault zone, North Vietnam Earth Sci Rev, 2012, 114, 224-235 Replumaz A., Lacassin R., Tapponnier P., Leloup P H., Large river offsets and Plio-Quaternary dextral strike-slip rate on the Red River fault (Yunnan, China) J Geophys Res., 2001, 106, 819–836 Roger F., Leloup P H., Jolivet M., Lacassin R., Phan Trong T., Brunel M., Seward D., Unravelling a long and complex thermal history by multi-system geochronology: example of the Song Chay metamorphic dome, North Vietnam Tectonophysics, 2000, 321, 449-466 Slemmons D B., Determination of design earthquake magnitude for micronation Proceedings “Third International Earthquake Microzonation Conference”; Seattle 1982, 1, 119-130 Schoenbohm L., Whipple K X., Burchfiel B C., River incision into a relict landscape along the Ailao Shan shear zone and Red River fault in Yunnan Province, China Geol Soc Am Bull., 2004, 116, 895–909 ❤tt♣✿✴✴❞①✳❞♦✐✳♦r❣✴✶✵✳✶✶✸✵✴❇✷✺✸✻✹✳✶ Schoenbohm L M., Whipple K X., Burchfiel B C., Chen L., Geomorphic constraints on surface uplift, exhumation, and plateau growth in the Red River region, Yunnan Province, China Geol Soc Am Bull., 2004, 116, 895–909 Tapponnier P., Lacassin R., Leloup P H., Scharer U., Zhong D., Liu X., Ji Sh., Zhang L., Zhong J., The Ailao Shan/Red River metamorphic belt: Tertiary leftlateral shear between Indochina and south China Nature ,1990, 343, 431-437 To T D., Yem N T., Cong D C., Hai V Q., Zuchiewicz W., Cuong N Q., Nghia N V., Recent crustal movements of northern Vietnam from GPS data J Geodynamics 2012, doi:10.1016/j.jog.2012.02.009 Well D L., Coppersmith K J., New empirical relationship among magnitude, rupture length, rupture width, rupture area and surface displacement Bulletin of the Seismological Society of America, 1994, 84, 974-1002 Winter Th., Phan Trong T., Lacassin R., Yem N T., [28] [29] [30] [31] [32] [33] [34] [35] Costaz J., Advantage of a deterministic approach of seismic risk for dam design : the Hoa-Binh dam case (Vietnam) Proceedings “Seismic Hazards in South East Asia” Hanoi, UNESCO, 1994, 249-254 Woodward-Clyde Consultants Seismic exposure study, offshore, southern California Report to Texaco USA, New Orleans, 1983 Wyss M., Estimating maximum expectable magnitude of earthquake from fault dimension Geology, 1979, 7, 336-340 Wang E., Burchfiel B C., Interpretation of Cenozoic tectonics in the right-lateral accommodation zone between the Ailao Shan shear zone and the Eastern Himala-yan syntaxis International Geology Review, 1997, 39, 191–219 Wang E., Burchfiel B C., Late Cenozoic to Holocene deformation in southwestern Sichuan and adjacent Yunnan, China, and its role in formation of the southeastern part of the Tibetan Plateau, Bulletin of the Seismological Society of America, 2000, 112, 413– 423 Weldon R., Sieh K., Zhu C., Han Y., Yang J., Robinson S., Slip rate and recurrence interval of earthquake on the Hong He (Red River) Fault, Yunnan., International Workshop on Seismotectonics and Seismic Hazard in SE Asia,1994, 244–248 Xiang J., Dong G., The strong ground motion records obtained in Langcang - Gengma earthquake in 1988, China, and their application in Proceedings “ Seismic Hazards in South East Asia” Hanoi , UNESCO, 1994, 94-99 Zuchiewicz W., Cuong N Q., Zasadni J , Yem N T., Late Cenozoic tectonics of the Red River Fault Zone, Vietnam, in the light of geomorphic studies, J Geodynamics, doi: ttp://elib.isivast.org.vn:2070/10.1016/j.jog.2011.10.008 Zuchiewicz W., Cuong N Q., Bluszcz, A., Michalik M., Quaternary sediments in the Dien Bien Phu fault zone, NW Vietnam: a record of young tectonic processes in the light of OSL-SAR dating results Geomorphology , 2004, 60, 269–302 233 Active fault segmentation and seismic hazard in Hoa-Binh reservoir, Vietnam Appendix A Table No Table Seismic data 1227-2003 of the Hoa Binh area and surrounding region No Year Month Day longitude latitude Depth M 1277 1285 1278 21 105.85 15 21 105.85 10 22 28 1928 12 21.42 41 1934 1 21.17 32 1931 45 1936 24 10 20.8 21.25 105.17 12 4.8 105.25 105.25 20.75 105.25 20.75 105.33 65 1945 73 1951 75 1952 88 1958 95 1962 97 1963 9 22 21 119 1967 129 1974 129 1974 142 1976 147 1976 486 1989 493 1989 491 1989 490 1989 492 1989 104.97 13 104.9 105.5 104.83 105.8 4.7 10 20 10 15 4.4 5.3 4.1 2.6 21 105.66 20 20.63 105.1 17 21.34 105.31 20 21.34 105.31 20 30 20.66 104.88 17 28 6 6 1 3 4 5 5 21.35 21 20.63 20.84 20.93 105.9 105.9 105.66 105.1 105.47 104.82 4.1 21.4 105 11 4.8 2.8 487 1989 15 4.4 252 1983 473 1988 105.08 4.8 105.9 17 406 1984 15 21.35 12 253 1983 105.9 4.2 230 1980 388 1983 21.37 20 254 1983 20.83 142 1976 21 21.25 115 1966 119 1967 20.97 20 115 1966 21.08 102 1964 101 1964 5.1 104.92 105.7 12 66 1945 5.1 4.3 20.5 10 4.8 58 1942 21.25 105.85 12 21 52 1937 54 1939 234 20 17 21 20.82 105.44 14 14 20.63 2.3 3.1 2.6 18 2.7 3.6 10 3.7 20.79 105.26 16 3.9 22 20.8 22 2.3 105.21 20.71 14 4.1 20.7 14 26 104.97 4.1 3.2 15 104.95 5 104.95 20.51 2.8 2.5 20.55 105.39 3.6 20.65 3.7 20.8 20.78 105.23 20 105.3 15 105.4 105.36 (Continued) Seismic data 1227-2003 of the Hoa Binh area and surrounding region Year Month Day longitude latitude Depth M 501 1990 22 20.57 105.12 13 2.7 533 1991 14 21.11 104.79 2.3 19 2.2 526 1991 524 1991 545 1991 11 598 1995 630 1995 631 1995 618 1995 617 1995 11 11 6 706 1997 11 686 1997 670 1997 10 12 12 28 28 1 17 21.03 20.73 20.85 20.77 20.91 20.72 13 105.18 20 105.08 105.32 105.08 105.09 105.55 105.59 2.2 3.7 2.4 0.7 21.08 104.79 19 3.2 21.09 104.82 3.9 10 20.65 24 21.11 104.84 15 3.5 24 21.25 104.87 10 1.8 705 1997 11 10 21.23 105.54 2.4 702 1997 10 24 20.98 105.65 2.9 21.15 104.88 19 2.7 695 1997 683 1997 696 1997 690 1997 701 1997 722 1998 720 1998 728 1998 711 1998 712 1998 716 1998 752 1999 757 2000 813 2001 10 7 11 21.24 21.4 20.68 105.21 21 1.7 10 20.5 105.86 0.9 13 20.75 104.93 18 3.3 1 21.21 29 21.09 21.11 11 30 20.79 854 2001 11 860 2001 819 2001 977 2002 952 2002 957 2002 959 2002 11 11 11 12 23 25 104.81 19 105.38 104.86 2.2 2.5 2.2 105.08 16 1.8 20.8 105.11 1.6 20.79 105.3 105.1 20.79 105.31 0.7 20.79 105.32 1.7 30 20 105.3 20.77 20 0.7 0.9 1.2 16 21.14 105.78 10 2.9 21.16 105.42 3.6 105.67 10 2.3 105.09 10 1.5 7 11 20.71 1060 2003 29 20.73 20.54 4.9 20.66 16 23 1062 2003 3.3 19 1075 2003 21.23 105.59 17 1.6 21.36 104.8 18 15 859 2001 858 2001 105.61 21 2.1 20.79 855 2001 105.52 12 3.7 21.07 20 856 2001 104.87 10 10 816 2001 104.8 1086 2003 20.73 104.79 3.8 3.8 20.97 105.17 105.53 21.09 104.81 20.83 105.12 20.95 105.15 19 12 10 10 2.5 1.9 2.8 2.1 1.8 Phan Trong Trinh et al Table No (Continued) Seismic data 1227-2003 of the Hoa Binh area and surrounding region Year Month Day longitude latitude Depth M 1087 2003 27 20.88 105.16 1084 2003 12 20.53 105.21 1079 2003 1081 2003 1056 2003 1076 2003 1080 2003 6 6 17 19 25 12 17 20.65 20.62 20.53 20.9 20.68 105.18 12 1.2 105.23 10 1.8 105.24 10 1.3 105.24 105.26 10 10 1077 2003 0.6 15 20.95 105.33 10 1082 2003 27 20.86 105.4 10 1054 2003 1059 2003 1072 2003 4 21 29 31 20.95 20.84 20.98 10 105.34 105.5 105.66 10 10 3.1 2.3 1.2 2.4 3.2 235 ... observation in the field can reveal if it is active or old fault Fault scarp that cuts across Holocene fan 225 Active fault segmentation and seismic hazard in Hoa-Binh reservoir, Vietnam is a candidate... MCE of Hoa-Binh fault segments in the Hoa-Binh dam is presented in the Table MCE of HoaBinh fault segment is 5.6 PGA caused by Hoa-Binh fault segment at the Hoa-Binh dam is represented in the... of 950 and 475 years are represented in the Table and the Table 5, respectively 229 Active fault segmentation and seismic hazard in Hoa-Binh reservoir, Vietnam Table Earthquake occurred in the